First, some background. AI cloud platforms provide artificial intelligence (AI) and machine learning (ML) capabilities that are packaged to simplify the process of analyzing data: finding patterns, building models, performing simulations and what-if scenarios. The platforms also seek to broaden the scope of this technology to make it available to business analysts, software developments, enterprise architects and IT operations teams.
AI/ML platform vendors are offering platform capabilities and tools for many roles within an enterprise so that teams can develop, operationalize, and manage a growing portfolio of AI solutions,” said Mike Gualtieri, an analyst at Forrester Research.
With that background established, let’s explore these two leading AI cloud platforms, DataRobot and H2O.ai. We’ll review the key features and compare them to help your business select the optimum choice for its needs.
Also see: What is Artificial Intelligence
DataRobot vs. H2O.ai: Key Features
The H2O AI Cloud is designed to solve complex business problems via automated machine learning (autoML) capabilities that are said to transform how AI is created and consumed. It seeks to democratize AI while continuing to provide data scientists with the features they need.
The platform encompasses the data science lifecycle, optimizing workflows to increase the quantity and quality of projects delivered to business stakeholders. Features are included for machine learning interpretability, and operations. It is used by data scientists, developers, machine learning engineers, analysts, and IT Professionals.
H2O.ai provides high performance computing with Nvidia Rapids integration and can scale workloads with support for network acceleration of Ampere-based Nvidia GPUs and the latest CUDA runtime. Users can monitor system usage and provide metrics for resource monitoring and autoscaling of multi-node clusters through a platform performance API. And its AI AppStore helps business users find and access custom AI applications.
DataRobot AI Cloud is a machine learning platform for automating, assuring, and accelerating predictive analytics. It helps data scientists and analysts to rapidly build and deploy accurate predictive models in a fraction of the time required by other solutions. It can visually track prediction progress with charts and manage workloads and delays. This enables users to view which predictions are delayed, why they are delayed, and the time frame.
Another feature enables users to compare data drift across multiple features (or groups of features) as well multiple time periods. This can be done for both training and scoring data. Available in the cloud, there is also a managed version known as DataRobot Dedicated Managed AI Cloud. This dedicated hosted version of AI Cloud is managed by DataRobot experts to support AI and machine learning projects with the advantage of public cloud services, reducing cost and time-to-value in deploying, upgrading and managing the AI infrastructure.
DataRobot vs. H2O.ai: Usability and Support
H2O.ai includes several features aimed at usability. It can automatically visualize and address data quality issues that transform your data into an optimal modeling dataset. It can visually display statistical properties within datasets and expose unexpected data quality issues like outliers, correlations, or missing values. Visual and text descriptions automatically help detect trends and insights including topics in text, correlations, and outliers.
Users, too, seem to appreciate H2O.ai’s no code approach that accelerates the process of building complex ML models. This is said to make it possible to create detailed models in minutes. It is well integrated with Amazon S3, Microsoft SQL Server database and other tools to provides a decent level of security, according to users. On the downside, users say it is still playing catchup with more established vendors like DataRobot and lacks some capabilities. One user said it was hard to modify data once uploaded and that cross-user access is limited. Others commented on a learning curve for deployment and dealing with expert settings aimed at data scientists. Customer support and documentation are also works in progress.
DataRobot AI Cloud comes with plenty of automation bells and whistles to speed up mundane data science and model creation operations. It can run on any combination of public clouds, data centers, or at the edge. Users like its governance and data protection capabilities.
DataRobot AI Cloud’s single-platform approach serves data scientists, analytics experts, IT, and the business with a single view of all data from any source, any type. Users comment that it is simple to use and adapt to your own needs. It is also said to be quick and effective at building models of most types.
Automation eliminates much of the coding work. It automates model setup, testing, and administration. Models can be built from any data source including tables and plain text, as well as graphic and geographical data. But there are a few negative comments, too. Sometimes the tools to decipher results are not the most user friendly, and there can be a lack of flexibility. Some say it favors modelling quantity over quality at times. And others complain that there are so many options that it can be difficult to know what to choose. One user said he wanted it to be easier to upload and integrate massive datasets.
In terms of usability and support, there is little to differentiate the two solutions. H2O.ai gets the nod due to a slightly higher ease of use rating from Gartner Peer Reviews.
DataRobot vs. H2O.ai: Data Scientists and Developers
Both platforms are used by data scientists. H20.ai has advanced autoML capabilities that encompass the entire data science lifecycle, optimizing workflows to increase both the quantity and quality of data science projects delivered to business stakeholders.
It contains robust features for machine learning interpretability and operations, and offers control over the entire explainability pipeline. For developers, H2O.ai automatically converts deployment artifacts from Python/R to Java/C++ to simplify the handoff of projects between data science teams and developers. It takes an open-source, low-code approach to development with enough flexibility to work with all major libraries and frameworks. Model monitoring features support for both H2O.ai and third-party models. They are accessible through a visual interface or API.
DataRobot also makes it easy to use open-source modeling techniques from R, Python, Spark, H2O, VW, XGBoost, and more. Small data science teams can use it to build and deploy a great many models. Instead of having to laboriously create each one, it acts as a productivity multiplier. Its library of algorithms and pre-built prototypes are useful for feature extraction and data preparation. Automation makes it possible to select and combine multiple algorithms to produce more accurate predictive models. It incorporates data science best practices. The platform enables analysts and others to perform data modeling tasks that previously required a high level of skill.
There is little to choose between both platforms in this category.
Also see: The Future of Artificial Intelligence
DataRobot vs. H2O.ai: Analyst View
The most recent Forrester Research analysis of AI/ML platforms places DataRobot well ahead of Leader H2O.ai. Forrester named DataRobot as one of three companies to earn a Leader designation. The analyst firm said its customers “appreciate the company’s rise from a niche automated machine learning (AutoML) player to a full-lifecycle AI platform.”
Strengths are listed as tooling and functionality in data preparation, model evaluation and explanation, ModelOps, and application building. It has gained ground recently by being able to augment and enrich data for modelling and adding a plug-in framework to make it easy for partners to add platform capabilities. Forrester said it has a relatively low barrier to entry for those adopting the platform and users commented on its ease of use and model documentation.
“DataRobot is a solid option for enterprises that want a platform that has tooling for extended AI teams while simultaneously providing collaboration and scale to manage existing use cases and crank out new one,” said Forrester’s Gualtieri.
H2O.ai earned a Contender rating from Forrester. The analyst firm liked its approach to ML but said it needed more of an AI platform focus. Its open-source ML algorithms leverage distributed computing architectures and automation that data scientists can use. But it lacks tooling for extended AI teams that would make the platform more attractive to technology partners. “H2O.ai has strengths in training, inferencing, and ModelOps. AI teams using Driverless AI can be significantly more productive than straight code-first data science teams,” said Gualtieri. “Areas of improvement include data management, security, and pluggable architecture for partners.”
On Gartner Peer Reviews, DataRobot also came out ahead. Users scored DataRobot higher on data exploration, visualization data preparation, model management, and collaboration. H2O.ai scored higher on forecasting, optimization, simulation, and user experience.
DataRobot wins in this category.
DataRobot vs. H20.ai: Conclusion
Pricing of both platforms is far from open. But the consensus is that DataRobot works out less expensive overall. Analyst ratings, user comments, and feature set appear to favor DataRobot over H2O.ai.
But H2O.ai is rising fast. It is steadily adding features as it scales from an ML algorithm specialist into becoming more of a full-featured AI platform. DataRobot has already made that journey so remains ahead. But for how long?
Also see: Top Data Mining Tools